Hacker News new | past | comments | ask | show | jobs | submit login

Do perfectly rational highly intelligent agents argue or is it our imperfection that needs such a tool? If yes then can people really change their minds after discussing like this?

(I do think its a great platform)




> Do perfectly rational highly intelligent agents argue [...]?

They agree, and do so quickly. See http://www.scottaaronson.com/blog/?p=2410 for an explanation of Aumann’s Agreement Theorem.


I think that the answer to the first part of your question is 'yes'. Just imagine two (perfectly rational and highly intelligent) mathematicians debating the truth of an unknown result—say, the Riemann hypothesis. I suppose if you add the caveat 'omniscient', then the need to argue goes away.


They argue, but they agree surely and quickly. (See "Are Disagreements Honest?" https://mason.gmu.edu/~rhanson/deceive.pdf)


It's interesting, I thought about it.. It seems to me that agreement can break when you aren't forced to reveal your model of the world. (For example, you may not even agree to use Bayesian statistics or logic.)

I think it could be useful if we forced the opponent in the debate not just to present argument, but also to reveal their model of the world. That way, they couldn't just poke holes into the argument from an inconsistent position.


thank you for posting these links. they have provided me with mind expanding knowledge


I published this blog post after reading maybe you would find it interesting to disagree about :)


Which blog post? Perhaps you want to post a link to it?





Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: